arXiv 2601.156282026-01-22cs.AI

CogToM：一种受人类认知启发的大型语言模型全面理论思维基准

CogToM: A Comprehensive Theory of Mind Benchmark inspired by Human Cognition for Large Language Models

BrainCog Lab, Institute of Automation, Chinese Academy of Sciences（脑认知实验室，自动化研究所，中国科学院）；Beijing Institute of AI Safety and Governance (Beijing-AISI)（北京人工智能安全与治理研究所）；Beijing Key Laboratory of Safe AI and Superalignment（北京安全人工智能与超对齐重点实验室）；School of Artificial Intelligence, UCAS（人工智能学院，中国科学院大学）；Long-term AI（长期人工智能）

Haibo Tong, Zeyang Yue, Feifei Zhao, Erliang Lin, Lu Jia, Ruolin Chen, Yinqian Sun, Qian Zhang, Yi Zeng

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

CogToM通过8000个双语实例评估LLM的理论思维能力，揭示性能异质性和认知瓶颈，为研究LLM认知边界提供新视角。

摘要：

大型语言模型（LLMs）是否真正具备类人理论思维（ToM）能力已引起越来越多关注。然而，现有的基准测试仍然主要局限于狭隘的范式，如虚假信念任务，未能捕捉到人类认知机制的完整范围。我们引入CogToM，一个包含超过8000个双语实例的综合性、理论基础的基准测试，涵盖46种范式，并通过49名人类标注者验证。对22个代表性模型的系统评估，包括前沿模型如GPT-5.1和Qwen3-Max，揭示了显著的性能异质性，并突显了特定维度上的持续瓶颈。进一步基于人类认知模式的分析表明，LLM与人类认知结构之间可能存在差异。CogToM为研究LLM认知边界的发展提供了稳健的工具和视角。

英文摘要：

Whether Large Language Models (LLMs) truly possess human-like Theory of Mind (ToM) capabilities has garnered increasing attention. However, existing benchmarks remain largely restricted to narrow paradigms like false belief tasks, failing to capture the full spectrum of human cognitive mechanisms. We introduce CogToM, a comprehensive, theoretically grounded benchmark comprising over 8000 bilingual instances across 46 paradigms, validated by 49 human annotator.A systematic evaluation of 22 representative models, including frontier models like GPT-5.1 and Qwen3-Max, reveals significant performance heterogeneities and highlights persistent bottlenecks in specific dimensions. Further analysis based on human cognitive patterns suggests potential divergences between LLM and human cognitive structures. CogToM offers a robust instrument and perspective for investigating the evolving cognitive boundaries of LLMs.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。